1//! Types supporting the UTF-8 parser
2
3/// Action to take when receiving a byte
4#[derive(Debug, Copy, Clone)]
5pub enum Action {
6 /// Unexpected byte; sequence is invalid
7 InvalidSequence = 0,
8 /// Received valid 7-bit ASCII byte which can be directly emitted.
9 EmitByte = 1,
10 /// Set the bottom continuation byte
11 SetByte1 = 2,
12 /// Set the 2nd-from-last continuation byte
13 SetByte2 = 3,
14 /// Set the 2nd-from-last byte which is part of a two byte sequence
15 SetByte2Top = 4,
16 /// Set the 3rd-from-last continuation byte
17 SetByte3 = 5,
18 /// Set the 3rd-from-last byte which is part of a three byte sequence
19 SetByte3Top = 6,
20 /// Set the top byte of a four byte sequence.
21 SetByte4 = 7,
22}
23
24/// States the parser can be in.
25///
26/// There is a state for each initial input of the 3 and 4 byte sequences since
27/// the following bytes are subject to different conditions than a tail byte.
28#[allow(non_camel_case_types)]
29#[derive(Debug, Default, Copy, Clone, PartialEq, Eq)]
30pub enum State {
31 /// Ground state; expect anything
32 #[default]
33 Ground = 0,
34 /// 3 tail bytes
35 Tail3 = 1,
36 /// 2 tail bytes
37 Tail2 = 2,
38 /// 1 tail byte
39 Tail1 = 3,
40 /// UTF8-3 starting with E0
41 U3_2_e0 = 4,
42 /// UTF8-3 starting with ED
43 U3_2_ed = 5,
44 /// UTF8-4 starting with F0
45 Utf8_4_3_f0 = 6,
46 /// UTF8-4 starting with F4
47 Utf8_4_3_f4 = 7,
48}
49
50impl State {
51 /// Advance the parser state.
52 ///
53 /// This takes the current state and input byte into consideration, to determine the next state
54 /// and any action that should be taken.
55 #[inline]
56 pub fn advance(self, byte: u8) -> (State, Action) {
57 match self {
58 State::Ground => match byte {
59 0x00..=0x7f => (State::Ground, Action::EmitByte),
60 0xc2..=0xdf => (State::Tail1, Action::SetByte2Top),
61 0xe0 => (State::U3_2_e0, Action::SetByte3Top),
62 0xe1..=0xec => (State::Tail2, Action::SetByte3Top),
63 0xed => (State::U3_2_ed, Action::SetByte3Top),
64 0xee..=0xef => (State::Tail2, Action::SetByte3Top),
65 0xf0 => (State::Utf8_4_3_f0, Action::SetByte4),
66 0xf1..=0xf3 => (State::Tail3, Action::SetByte4),
67 0xf4 => (State::Utf8_4_3_f4, Action::SetByte4),
68 _ => (State::Ground, Action::InvalidSequence),
69 },
70 State::U3_2_e0 => match byte {
71 0xa0..=0xbf => (State::Tail1, Action::SetByte2),
72 _ => (State::Ground, Action::InvalidSequence),
73 },
74 State::U3_2_ed => match byte {
75 0x80..=0x9f => (State::Tail1, Action::SetByte2),
76 _ => (State::Ground, Action::InvalidSequence),
77 },
78 State::Utf8_4_3_f0 => match byte {
79 0x90..=0xbf => (State::Tail2, Action::SetByte3),
80 _ => (State::Ground, Action::InvalidSequence),
81 },
82 State::Utf8_4_3_f4 => match byte {
83 0x80..=0x8f => (State::Tail2, Action::SetByte3),
84 _ => (State::Ground, Action::InvalidSequence),
85 },
86 State::Tail3 => match byte {
87 0x80..=0xbf => (State::Tail2, Action::SetByte3),
88 _ => (State::Ground, Action::InvalidSequence),
89 },
90 State::Tail2 => match byte {
91 0x80..=0xbf => (State::Tail1, Action::SetByte2),
92 _ => (State::Ground, Action::InvalidSequence),
93 },
94 State::Tail1 => match byte {
95 0x80..=0xbf => (State::Ground, Action::SetByte1),
96 _ => (State::Ground, Action::InvalidSequence),
97 },
98 }
99 }
100}
101